Skip to content

Benchmark

CL-bench · Domain Knowledge Reasoning

Scores are percentages. Higher is better.

Models and systems measured
33
Scores recorded
58
Sources
2
IQ weight
1.59

Score distribution

19.1%
Mean
19.5%
Median
4.3%
Std dev
13.5%
10th percentile
23.9%
90th percentile
11.9%
Min
11.9%28.8%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.59 for chat models, measured over 20 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#Model or systemScore
1GPT 5.4benchmark-owner leaderboard run · site model-mode label: GPT-5.4 (xhigh) · 1899 tasks · GPT-5.1 judge · low reasoning effort · single run; no standard deviation · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs28.8%
2GPT 5.4xhigh· Epoch AI Benchmarking Hub data28.8%
3GPT 5.1high· Epoch AI Benchmarking Hub data25.3%
4Grok 4.20· Epoch AI Benchmarking Hub data23.9%
5Claude Opus 4.5· Epoch AI Benchmarking Hub data23.7%
6Claude Opus 4.5benchmark-owner leaderboard run · site model-mode label: Claude Opus 4.5 Thinking · 1899 tasks · GPT-5.1 judge · low reasoning effort · site standard deviation ±1.2 percentage points · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs23.7%
7GPT 5.1· Epoch AI Benchmarking Hub data22.4%
8Qwen3.6 Plus· Epoch AI Benchmarking Hub data22.2%
9Qwen3.6 Plusbenchmark-owner leaderboard run · site model-mode label: Qwen 3.6 Plus · 1899 tasks · GPT-5.1 judge · low reasoning effort · single run; no standard deviation · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs22.2%
10Qwen3.5 Plus· Epoch AI Benchmarking Hub data22%
11Gemini 3.1 Pro Preview· Epoch AI Benchmarking Hub data21.1%
12Gemini 3.1 Pro Previewbenchmark-owner leaderboard run · site model-mode label: Gemini 3.1 Pro (high) · 1899 tasks · GPT-5.1 judge · low reasoning effort · single run; no standard deviation · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs21.1%
13Claude Opus 4.5benchmark-owner leaderboard run · site model-mode label: Claude Opus 4.5 · 1899 tasks · GPT-5.1 judge · low reasoning effort · site standard deviation ±0.5 percentage points · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs21%
14GLM 5· Epoch AI Benchmarking Hub data20.8%
15GLM 5benchmark-owner leaderboard run · site model-mode label: GLM 5.0 · 1899 tasks · GPT-5.1 judge · low reasoning effort · single run; no standard deviation · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs20.8%
16Kimi K2.5· Epoch AI Benchmarking Hub data20.8%
17Kimi K2.5benchmark-owner leaderboard run · site model-mode label: Kimi K2.5 · 1899 tasks · GPT-5.1 judge · low reasoning effort · single run; no standard deviation · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs20.8%
18Claude Opus 4.6· Epoch AI Benchmarking Hub data20.4%
19Claude Opus 4.6benchmark-owner leaderboard run · site model-mode label: Claude Opus 4.6 · 1899 tasks · GPT-5.1 judge · low reasoning effort · single run; no standard deviation · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs20.4%
20GPT 5.2· Epoch AI Benchmarking Hub data19.5%
21GPT 5.2benchmark-owner leaderboard run · site model-mode label: GPT 5.2 · 1899 tasks · GPT-5.1 judge · low reasoning effort · site standard deviation ±1.5 percentage points · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs19.5%
22Kimi K2 Thinking· Epoch AI Benchmarking Hub data18.7%
23GPT 5.2benchmark-owner leaderboard run · site model-mode label: GPT 5.2 (High) · 1899 tasks · GPT-5.1 judge · low reasoning effort · site standard deviation ±0.9 percentage points · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs18.6%
24GPT 5.2high· Epoch AI Benchmarking Hub data18.6%
25OpenAI o3high· Epoch AI Benchmarking Hub data18%
26GLM 4.7· Epoch AI Benchmarking Hub data17.9%
27MiMo V2 Pro· Epoch AI Benchmarking Hub data17.7%
28MiMo V2 Probenchmark-owner leaderboard run · site model-mode label: MiMo V2 Pro · 1899 tasks · GPT-5.1 judge · low reasoning effort · single run; no standard deviation · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs17.7%
29Gemini 3 Pro Preview· Epoch AI Benchmarking Hub data15.5%
30Gemini 3 Pro Previewbenchmark-owner leaderboard run · site model-mode label: Gemini 3 Pro (High) · 1899 tasks · GPT-5.1 judge · low reasoning effort · site standard deviation ±1.1 percentage points · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs15.5%
31Qwen3 Max· Epoch AI Benchmarking Hub data14.6%
32Gemini 3 Pro Previewbenchmark-owner leaderboard run · site model-mode label: Gemini 3 Pro · 1899 tasks · GPT-5.1 judge · low reasoning effort · site standard deviation ±0.3 percentage points · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs13.9%
33DeepSeek V3.2benchmark-owner leaderboard run · site model-mode label: DeepSeek V3.2 Thinking · 1899 tasks · GPT-5.1 judge · low reasoning effort · site standard deviation ±0.6 percentage points · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs13.6%
34DeepSeek V3.2 Expthinking· Epoch AI Benchmarking Hub data13.6%
35DeepSeek V3.2· Epoch AI Benchmarking Hub data12.3%
36DeepSeek V3.2benchmark-owner leaderboard run · site model-mode label: DeepSeek V3.2 · 1899 tasks · GPT-5.1 judge · low reasoning effort · site standard deviation ±0.8 percentage points · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs12.3%
37MiniMax M2.5· Epoch AI Benchmarking Hub data12.2%
38MiniMax M2.5benchmark-owner leaderboard run · site model-mode label: MiniMax M2.5 · 1899 tasks · GPT-5.1 judge · low reasoning effort · single run; no standard deviation · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs12.2%

Not ranked

The source marks these results as descriptive, not capability scores, so we list them by name instead of ranking them.

Model or systemScore
Doubao 1.6 (CL-bench label)benchmark-owner leaderboard run · site model-mode label: Doubao 1.6 · 1899 tasks · GPT-5.1 judge · low reasoning effort · site standard deviation ±0.5 percentage points · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs11.9%
Doubao 1.6 (CL-bench label)benchmark-owner leaderboard run · site model-mode label: Doubao 1.6 Thinking · 1899 tasks · GPT-5.1 judge · low reasoning effort · site standard deviation ±0.1 percentage points · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs13.7%
GLM-4.7 (CL-bench label)benchmark-owner leaderboard run · site model-mode label: GLM 4.7 · 1899 tasks · GPT-5.1 judge · low reasoning effort · single run; no standard deviation · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs17.9%
GPT-5.1 (CL-bench label)benchmark-owner leaderboard run · site model-mode label: GPT 5.1 · 1899 tasks · GPT-5.1 judge · low reasoning effort · site standard deviation ±0.7 percentage points · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs22.4%
GPT-5.1 (CL-bench label)benchmark-owner leaderboard run · site model-mode label: GPT 5.1 (High) · 1899 tasks · GPT-5.1 judge · low reasoning effort · site standard deviation ±1.3 percentage points · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs25.3%
Grok 4.20benchmark-owner leaderboard run · site model-mode label: Grok 4.20 (non-reasoning) · 1899 tasks · GPT-5.1 judge · low reasoning effort · single run; no standard deviation · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs19.5%
Grok 4.20benchmark-owner leaderboard run · site model-mode label: Grok 4.20 (reasoning) · 1899 tasks · GPT-5.1 judge · low reasoning effort · single run; no standard deviation · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs23.9%
Hy2.0 (CL-bench label)benchmark-owner leaderboard run · site model-mode label: Hy2.0 · 1899 tasks · GPT-5.1 judge · low reasoning effort · site standard deviation ±0.9 percentage points · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.14.3%
Hy2.0 (CL-bench label)benchmark-owner leaderboard run · site model-mode label: Hy2.0 Thinking · 1899 tasks · GPT-5.1 judge · low reasoning effort · site standard deviation ±1 percentage points · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.18%
Hy3 Preview (CL-bench label)benchmark-owner leaderboard run · site model-mode label: Hy3 preview · 1899 tasks · GPT-5.1 judge · low reasoning effort · single run; no standard deviation · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.25.2%
Kimi K2 (CL-bench label)benchmark-owner leaderboard run · site model-mode label: Kimi K2 · 1899 tasks · GPT-5.1 judge · low reasoning effort · site standard deviation ±1.1 percentage points · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs12.3%
Kimi K2 (CL-bench label)benchmark-owner leaderboard run · site model-mode label: Kimi K2 Thinking · 1899 tasks · GPT-5.1 judge · low reasoning effort · site standard deviation ±0.6 percentage points · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs18.7%
o3 (CL-bench label)benchmark-owner leaderboard run · site model-mode label: o3 (High) · 1899 tasks · GPT-5.1 judge · low reasoning effort · site standard deviation ±1.4 percentage points · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs18%
Polyphony (CL-bench label)benchmark-owner leaderboard run · site model-mode label: Polyphony · 1899 tasks · GPT-5.1 judge · low reasoning effort · site standard deviation ±1 percentage points · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs25.1%
Qwen 3 Max (CL-bench label)benchmark-owner leaderboard run · site model-mode label: Qwen 3 Max · 1899 tasks · GPT-5.1 judge · low reasoning effort · site standard deviation ±1 percentage points · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs14.6%
Qwen 3 Max (CL-bench label)benchmark-owner leaderboard run · site model-mode label: Qwen 3 Max Thinking · 1899 tasks · GPT-5.1 judge · low reasoning effort · site standard deviation ±0.5 percentage points · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs13.5%
Qwen3.5 Plusbenchmark-owner leaderboard run · site model-mode label: Qwen 3.5 Plus (thinking) · 1899 tasks · GPT-5.1 judge · low reasoning effort · single run; no standard deviation · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs22%
Seed 1.8 (CL-bench label)benchmark-owner leaderboard run · site model-mode label: SEED 1.8 · 1899 tasks · GPT-5.1 judge · low reasoning effort · single run; no standard deviation · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs16.9%
Seed 2.0 (CL-bench label)benchmark-owner leaderboard run · site model-mode label: SEED 2.0 (high) · 1899 tasks · GPT-5.1 judge · low reasoning effort · single run; no standard deviation · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs21%
Seed 2.0 (CL-bench label)benchmark-owner leaderboard run · site model-mode label: SEED 2.0 (medium) · 1899 tasks · GPT-5.1 judge · low reasoning effort · single run; no standard deviation · candidate API snapshot and exact candidate harness not specified on public leaderboard· Reported by Tencent Hunyuan / Fudan NLP Group· via BenchBoard docs20.1%

Data credits: BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links. · Epoch AI Benchmarking Hub, CC BY 4.0 (licence)